www.ijecs.in
International Journal Of Engineering And Computer Science ISSN:2319-7242
Volume 5 Issue 8 August 2016 Page No. 17658-17661
A Review on Plant Leaf Classification and Segmentation
Lakhvir Kaur1, Dr. Vijay Laxmi2
¹Research Scholar M-Tech, Computer Science and Engineering, Guru Kashi University, India
²Dean, UCCA, Guru Kashi University, India
1
lakhvirgrewal92@gmail.com, 2cse_vijay2003@yahoo.co.in
Abstract: A leaf is an organ of vascular plant and is the principal lateral appendage of the stem. Each leaf has a set of features that
differentiate it from the other leaves, such as margin and shape. This paper proposes a comparison of supervised plant leaves
classification using different approaches, based on different representations of these leaves, and the chosen algorithm. Beginning with
the representation of leaves, we presented leaves by a fine-scale margin feature histogram, by a Centroid Contour Distance Curve
shape signature, or by an interior texture feature histogram in 64 element vector for each one, after we tried different combination
among these features to optimize results. We classified the obtained vectors. Then we evaluate the classification using cross
validation. The obtained results are very interesting and show the importance of each feature. the classification of plant leaf images
with biometric features. Traditionally, the trained taxonomic perform this process by following various tasks. The taxonomic usually
classify the plants based on flowering and associative phenomenon. It was found that this process was time consuming and difficult.
The biometric features of plants leaf like venation make this classification easy. Leaf biometric feature are analyzed using computer
based method like morphological feature analysis and artificial neural network based classifier. KNN model take input as the leaf
venation morphological feature and classify them into four different species.
Keyword - Artificial neural network, Canny edge detection, kNN Classification, Leaf venation pattern, Morphological
Features
I. INTRODUCTION
Plant diseases have turned into a dilemma as it can cause
significant reduction in both quality and quantity of agricultural
products. In India 70% of the population depend on agriculture.
Farmers have wide range of diversity to select suitable Fruit and
Vegetable crops. However, the cultivation of these crops for
optimum yield and quality produce is highly technical. It can be
improved by the aid of technological support. The management
of perennial fruit crops requires close monitoring especially for
the management of diseases that can affect production
significantly and subsequently the postharvest life. In [2] the
authors have worked on the development of methods for the
automatic classification of leaf diseases based on high resolution
multispectral and stereo images. Leaves of sugar beet are used
for evaluating their approach. Sugar beet leaves might be
infected by several diseases, such as rusts (Uromyces betae),
powdery mildew (Erysiphe betae).
Disease is caused by pathogen which is any agent causing
disease. In most of the cases pests or diseases are seen on the
leaves or stems of the plant. Therefore identification of plants,
leaves, stems and finding out the pest or diseases, percentage of
the pest or disease incidence , symptoms of the pest or disease
attack, plays a key role in successful cultivation of crops. It is
found that diseases cause heavy crop losses amounting to several
billion dollars annually.
Disease management is a challenging task. Mostly diseases are
seen on the leaves or stems of the plant. Precise quantification of
these visually observed diseases, pests, traits has not studied yet
because of the complexity of visual patterns. Hence there has
been increasing demand for more specific and sophisticated
image pattern understanding. In biological science, sometimes
thousands of images are generated in a single experiment. These
images can be required for further studies like classifying lesion,
scoring quantitative traits, calculating area eaten by insects, etc.
Almost all of these tasks are processed manually or with distinct
software packages. It is not only tremendous amount of work but
also suffers from two major issues: excessive processing time
and subjectiveness rising from different individuals .Hence to
conduct high throughput experiments, plant biologist need
efficient computer software to automatically extract and analyze
significant content. Here image processing plays important role.
Lakhvir Kaur1, IJECS Volume 05 Issue 08 August 2016 Page No.17658-17661
Page 17658
DOI: 10.18535/ijecs/v5i8.41
II. IMPORTANCE
In object recognition research a lot have been done about
general features extraction or recognition between different
classes of objects. In case of a species domain recognition, 10
taking into account the unique characteristics that belong to this
category, improves the performance of the system. Despite the
high technical aspect of this project, dealing with leaves, gives it
a biological connotation. Some basic knowledge about leaves
have to be learned and concepts about how the biologists
themselves recognize leaves has to be studied. The next two
paragraphs are devoted to these experiences. Precision Botany
(PB) refers to the application of new technologies in plant
identification. Computer vision can be used in PB to distinguish
plants from its species level, so that an identification can be
applied on the size and number of plants detected for the
classification purpose. This is focused on the application of
computer vision for identification purposes of species in
Stemonoporus genus. Surveys reveal that there are 3711
flowering plant species in Sri Lanka [4]. Out of these, 926 are
endemic [5, 6]. Since some of these have minute variations,
identification of these species has become difficult. Accurate and
speedy identification of plants has become a time consuming and
a fuzzy work due to nonavailability of a computerized scientific
plant identification system. Design and implementation of
image-based plant classification system is a long felt. Biologists
receive a large number of requests to identify plants for people,
many species of plants look very similar on their leaves, and
botanists will turn to identifying the species based on their
structure or other morphologies.
There are three main parts to a leaf:
1. The base which is the point at which the leaf is joined to the
stem.
2. The stalk or petiole is the thin section joining the base to the
lamina - it is generally cylindrical or semicircular in form.
3. The lamina or leaf blade is the wide part of the leaf Leaves
can be of many different shapes: Primarily, leaves are divided
into simple - a single leaf blade with a bud at the base of the
leafstem; or compound - a leaf with more than one blade.
All blades are attached to a single leafstem. Where the leafstem
attaches to the twig there is a bud. Leaves may be arranged on
the stem either in an alternate arrangement – leaves that are not
places directly across from each other on the twig; or in an
opposite arrangement – 2 or 3 leaves that are directly across
from each other on the same twig.
The margin (the edge of a leaf) as seen in Figure 3.2 may be
entire, singlytoothed, doubly-toothed, or lobed. Compound
leaves may be palmate – having the leaflets arranged round a
single point like fingers On the palm of a hand; or pinnate –
when the leaves are joined on the two sides of the stalk, like the
vanes of a feather as seen in the Figure 1.3. Leaf arrangements
are pretty straightforward to figure out. Need to look for the
nodes and then determine how many leaves are coming off each
node. If there’s only one leaf per node, then need only to
determine whether the arrangement is alternate or spiral, and it’s
usually pretty obvious.
Figure 1. Simple leaves – Margin structure.
III. LITERATURE SURVEY
Smita Naikwadi et.al.[2013] We propose and experimentally
evaluate a software solution for automatic detection and
classification of plant leaf diseases. Studies of plant trait/disease
refer to the studies of visually observable patterns of a particular
plant. Nowadays crops face many traits/diseases. Damage of the
insect is one of the major trait/disease. Insecticides are not
always proved efficient because insecticides may be toxic to
some kind of birds. It also damages natural animal food chains.
The following two steps are added successively after th
segmentation phase. In the first step we identify the mostly green
colored pixels. Next, these pixels are masked based on specific
threshold values that are computed using Otsu's method, then
those mostly green pixels are masked. The other additional step
is that the pixels with zeros red, green and blue values and the
pixels on the boundaries of the infected cluster (object) were
completely removed. The experimental results demonstrate that
the proposed technique is a robust technique for the detection of
plant leaves diseases. The developed algorithm’s efficiency can
successfully detect and classify the examined diseases with a
precision between 83% and 94%, and can achieve 20% speedup
over the approach proposed in [1].
Abdul Kadir et.al. [2012]This paper reports the results of
experiments in improving performance of leaf identification
system using Principal Component Analysis (PCA). The system
involved combination of features derived from shape, vein,
color, and texture of leaf. PCA was incorporated to the
identification system to convert the features into orthogonal
features and then the results were inputted to the classifier that
used Probabilistic Neural Network (PNN). This approach has
been tested on two datasets, Foliage and Flavia, that contain
various color leaves (foliage plants) and green leaves
respectively. The results showed that PCA can increase the
accuracy of the leaf identification system on both datasets.[2]
Samuel E. Buttrey et.al. [2002] We construct a
hybrid(composite) classi&er by combining two classi&ers in
common use— classi&cation trees and k-nearest-neighbor (kNN). In our scheme we divide the feature space up by a
classi&cation tree, andthen classify test set items using the k-NN
rule just among those training items in the same leaf as the test
item.
This
reduces
somewhat
the
computational
loadassociatedwith k-NN, andit produces a classi&cation rule
that performs better than either trees or the usual k-NN in a
number of well-known data sets. [3]
Prof. Meeta Kumar et al. [2013]In this paper we present survey
on various classification techniques which can be used for plant
leaf classification. A classification problem deals with
Lakhvir Kaur1, IJECS Volume 05 Issue 08 August 2016 Page No.17658-17661
Page 17659
DOI: 10.18535/ijecs/v5i8.41
associating a given input pattern with one of the distinct classes.
Plant leaf classification is a technique where leaf is classified
based on its different morphological features. There are various
successful classification techniques like kNearest Neighbor
Classifier, Probabilistic Neural Network, Genetic Algorithm,
Support Vector Machine, and Principal Component Analalysis.
Deciding on the method for classification is often a difficult task
because the quality of the results can be different for different
input data. Plant leaf classifications has wide applications in
various fields such as botany, Ayurveda, Agriculture etc. The
goal of this survey is to provide an overview of different
classification techniques for plant leaf classification.[4]
Gurpreet kaur et.al.[2012] Plants play an important role in
human life and provide required information for the
development of human society. The urgent situation is that due
to environmental degradation and lack of awareness, many rare
plant species are at the risk of extinction so it is necessary to
keep record for plant protection. We believe that the first step is
to buildup a database for protecting plants. So the need arises to
teach a computer how to classify plants. Despite the great
advances made in Botany, there are many plants which are still
unknown.This research focuses on using digital image
processing for the purpose of automate classification and
recognition of plants based on the images of the leaves. In this
paper we review leaf architecture and various techniques for
automated plant classification and recognition. [5]
Comparison Table :
Methods Advantages
of
classifica
tion
K-NN
1. Simpler classifier
as exclusion of any
training
process.
2. Applicable in
case of a small
dataset which is not
trained.
SVM
1. Simple geometric
interpretation and a
sparse
solution.
2. Can be robust,
even when training
sample has some
bias.
ANN
1. Tolerant of noisy
inputs.
2.
Instances
classified by more
than one output.
1.
Easy
to
implement.
2. Applicable to
wide
range
of
problems.
Backprop
agation
Network
(BPN)
Disadvantages
1. Speed of computing
distance
increases
according to numbers
available in training
samples.
2. Expensive testing of
each
instance
and
sensitive to irrelevant
inputs.
1.
Slow
training.
2.
Difficult
to
understand structure of
algorithm.
3. Large no. support
vectors
are
needed from training set
to perform classification
task
1. Long training time.
2. Large complexity of
network
structure.
3. Need lot of memory
for training data
1. Learning can be slow.
2. It is hard to know how
many neurons as well as
layers are required.
3. Able to form
arbitrarily complex
nonlinear mappings
Radial
Basis
Function
(RBF) or
Naive
Bayesian
1. Training phase is
faster.
2. Hidden layer is
easier to interpret.
1. It is slower in
execution when speed is
a factor.
IV. PROBLEM STATEMENT
Plant leaf classification is the major problem and also their
diseases are also major problem. Most of plant leaf have attacks
by snail, worm and fungi. Furthermore, when the leaf had been
infected or attacked, the others areas had been exposed to be
infected. Thus, it will decrease the life of plant. There is the
overlapping of the plant diseases problem. Plants play an
important role in our lives, without plants there will not be the
existence of the ecology of the earth. The large amount of leaf
types now makes the human being in a front of some problems
in the specification of the use of plants, the first need to know
the use of a plant is the identification of the plant leaf. The
problem with a number of these techniques is that they require
some manual intervention such as correctly orienting the image
or identifying the end points of the leaf’s main vein. S. R.
Deokar (2013) has worked on leaf recognition by extracting 28
and 60 Feature point. These features extracted by vertical and
horizontal splitting the leaf images. ANN is used to compare
performance of leaf recognition.
•
•

V. SCOPE
The users of this system are paddy farmers.
The prototype will be develop by using MATLAB 2013
a.
10 samples each of the normal, brown spot disease,
narrow brown spot disease and blast disease will be
used in this project.
VI. CONCLUSION
Plants play an important role in our lives, without plants there
will not be the existence of the ecology of the earth. The large
amount of leaf types now makes the human being in a front of
some problems in the specification of the use of plants, the first
need to know the use of a plant is the identification of the plant
leaf. This work proposed a comparison of supervised
classification of plant leaves, where we used to represent species
in seven different representations, using three features extracted
from binary masks of these leaves: a finescale margin feature
histogram, by a Centroid Contour Distance Curve shape
signature, and by an interior texture feature histogram. Results
were very interesting in a way that gives as clear ideas: In term
of representation: we can differentiate leaves by its margin better
than shape or texture, but, experiments shown in this study prove
our idea: the more we combine these features, the more precise
the difference between samples is and that is what gives better
results in classification. In term of classification: distance based
algorithms give the best result for plant leaves classification. So,
we can conclude that these algorithms are the most suitable for
that task. On the other hand, the approach based on decision tree
Lakhvir Kaur1, IJECS Volume 05 Issue 08 August 2016 Page No.17658-17661
Page 17660
DOI: 10.18535/ijecs/v5i8.41
gives the worst results because of the overfitting problem. In
general, a learning algorithm is said to overfit relative to a
simpler one if it is more accurate in fitting known data but less
accurate in predicting new data. Use of the three features proved
that there is some information more important than other. We
discovered that margin representation can affect results more
than the shape of the leaf. However, the combination of the three
features gives the best result. To solve this problem, we plan, as
future work, use of feature extraction algorithms, like PSO, to
clean dataset and keep the important information in order to
optimize the obtained results and avoid overfitting problem
posed by decision tree algorithm. We plan also to use bioinspired algorithms. They are part of a new research domain that
is becoming more important due to its results in different areas.
REFERENCES
[1]. Smita Naikwadi et.al.―ADVANCES IN IMAGE PROCESSING
FOR DETECTION OF PLANT DISEASES‖ International Journal of
Application or Innovation in Engineering & Management (IJAIEM)
Web
Site:
www.ijaiem.org
Email:
editor@ijaiem.org,
editorijaiem@gmail.com Volume 2, Issue 11, November 2013.
[11]. C. Mallah, J. Cope, and J. Orwell ,"Plant leaf classification using
probabilistic integration of shape, texture and margin features," Signal
Processing, Pattern Recognition and Applications, 2013
[12]. S. Zhang, and K. W. Chau, "Dimension reduction using
semisupervised locally linear embedding for plant leaf classification,"
Emerging Intelligent Computing Technology and Applications,
Springer Berlin Heidelberg, 2009, pp. 948-955
[13]. A. Kadir, L. E. Nugroho, A. Susanto, and P. I. Santosa, "Leaf
classification using shape, color, and texture features , arXiv preprint
arXiv:1401.4447, 2013
[14]. A. H. M. Amin, and A. I. Khan, "One-shot Classification of 2-D
Leaf Shapes Using Distributed Hierarchical Graph Neuron (DHGN)
Scheme with k-NN Classifier," Procedia Computer Science, 24, 2013,
pp.84-96
[2]. Abdul Kadir et.al. ―Performance Improvement of Leaf Identification
System Using Principal Component Analysis‖ International Journal of
Advanced Science and Technology Vol. 44, July, 2012
[3]. Samuel E. Buttrey et.al. ―Using k-nearest-neighbor classi&cation in
the leaves of a tree‖ Computational Statistics & Data Analysis 40
(2002) 27 – 37 www.elsevier.com/locate/csda.
[4]. Prof. Meeta Kumar et al. ―Survey on Techniques for Plant Leaf
Classification‖ International Journal of Modern Engineering Research
(IJMER) www.ijmer.com Vol.1, Issue.2, pp-538-544 ISSN: 2249-6645.
[5]. Gurpreet kaur et.al. ―Classification of Biological Species Based on
Leaf Architecture–A review‖ IRACST - International Journal of
Computer Science and Information Technology & Security (IJCSITS),
ISSN: 2249-9555 Vol. 2, No.2, April 2012.
[6]. Amlekar Manisha, Manza R.R, Yannawar Pravin,(2013), Leaf
classification based on leaf dimension biometric features of leaf shape
using k-means classifier, NCAC, Jalgoan.
[7]. Amlekar Manisha, Manza R.R, Yannawar Pravin, Gaikwad
B.P,(2013), Image data mining for classifying leaf dimension biometric
features of leaf shape using KNN classification technique, CMS,
[8]. Aurangabad. Beghin T., Cope J. S., Remagnino P. and Barman S.,
(2010), Shape and texture based plant leaf Classification, ACIVS, 2,
345–353.
[9]. J. S., Remagnino P., Barman S., and Wilkin P., (2010), Plant texture
classification using gabor co-occurrences,‖ in Proceedings of the 6 th
international conference on Advances in visual computing,2, 669– 677.
[10]. Deokar S. R., Zope P. H. Suralkar S. R ,(2013), Leaf Recognition
Using Feature Point Extraction and Artificial Neural Network,
International Journal of Engineering Research & Technology (IJERT),
ISSN: 2278-0181, 2(1). Ehsanirad, (2010), Plant classification based on
leaf recognition, International Journal of Computer Science and
Information Security, 8(4), 78–81.
Lakhvir Kaur1, IJECS Volume 05 Issue 08 August 2016 Page No.17658-17661
Page 17661